Ethnic Differences in the Prevalence and Risk

Factors of Diabetic Retinopathy

The Singapore Epidemiology of Eye Diseases Study
Gavin S. Tan, MD, FRCSEd,1,2,3 Alfred Gan, MSc,1 Charumathi Sabanayagam, PhD,1,2,3
Yih Chung Tham, PhD,1 Kumari Neelam, MD,1,4 Paul Mitchell, MD, PhD,5 Jie Jin Wang, PhD,3
Ecosse L. Lamoureux, PhD,1,3 Ching-Yu Cheng, MD, PhD,1,2,3 Tien Y. Wong, MD, PhD1,2,3

Purpose: To evaluate the prevalence and risk factors for diabetic retinopathy (DR) in the Singapore Epide-
miology of Eye Diseases (SEED) Study.
Design: Population-based, cross-sectional study.
Participants: Persons of Malay, Indian, and Chinese ethnicity aged 40þ years, living in Singapore.
Methods: Diabetes was defined as nonfasting plasma glucose 200 mg/dl (11.1 mmol/l), glycated hemo-
globin A1c (HbA1c) >6.5%, self-reported physician-diagnosed diabetes, or the use of glucose-lowering medi-
cation. Retinal photographs, were graded for the presence and severity of DR using the modified Airlie House
classification system.
Main Outcome Measures: Diabetic retinopathy, diabetic macular edema (DME), vision-threatening diabetic
retinopathy (VTDR), defined as the presence of severe nonproliferative or proliferative DR, or clinically significant
macular edema (CSME).
Results: Of the 10 033 subjects, 2877 (28.7%) had diabetes and gradable photographs for analysis. The
overall age-standardized prevalence (95% confidence interval [CI]) was 28.2% (25.9e30.6) for any DR, 7.6%
(6.5e9.0) for DME, and 7.7% (6.6e9.0) for VTDR. Indians had a higher prevalence of any DR (30.7% vs. 26.2% in
Chinese and 25.5% in Malays, P ¼ 0.012); a similar trend was noted for any DME (P ¼ 0.001) and CSME (P ¼
0.032). Independent risk factors for any DR were Indian ethnicity (odds ratio [OR], 1.41; 95% CI, 1.09e1.83, vs.
Chinese), diabetes duration (OR, 1.10; 95% CI, 1.08e1.11, per year), HbA1c (OR, 1.25; 95% CI, 1.18e1.32, per
%), serum glucose (OR, 1.03; 95% CI, 1.00e1.06, per mmol/l), and systolic blood pressure (OR, 1.14; 95% CI,
1.09e1.19, per 10 mmHg). Diastolic blood pressure (OR, 0.74; 95% CI, 0.65e0.84, per 10 mmHg increase), total
cholesterol (OR, 0.87; 95% CI, 0.80e0.95, per mmol/l increase), and low-density lipoprotein (LDL) cholesterol
(OR, 0.83; 95% CI, 0.74e0.92, per mmol/l increase) were associated with lower odds of any DR. Risk factors were
largely similar across the 3 ethnic groups.
Conclusions: Indian Singaporeans have a higher prevalence of DR and DME compared with Chinese and
Malays. Major risk factors for DR in this study were similar across the 3 ethnic groups. Addressing these risk
factors may reduce the impact of DR in Asia, regardless of ethnicity. Ophthalmology 2017;-:1e8 ª 2017 by the
American Academy of Ophthalmology

Diabetic retinopathy (DR) is the most common ocular
complication of diabetes mellitus and the leading cause of
visual loss in working-age adults in the developed world.1e3
There is an increasing prevalence of diabetes in Asia,
particularly in India and China.4e7 The global prevalence of
DR among persons with diabetes is estimated at 34.6%, with
10.2% having vision-threatening levels of DR, although the
rates vary widely between countries and ethnic groups
globally.8
Previous work has observed racial/ethnic variation in the
prevalence of DR.9e11 For example, in Western pop-
ulations, ethnic blacks and Hispanics have a higher preva-
lence of DR and diabetic macular edema (DME) than
whites.10,12e14 These variations may reflect a combination
of disparities in socioeconomic status and healthcare access,
and a differential contribution of traditional risk factors for
DR such as poor glycemic blood pressure control and
duration of diabetes. Few studies have examined whether
ethnicity is an independent risk factor for DR. Although it
has been reported that there is no ethnic difference in the
association of glycated hemoglobin with retinopathy,15
nevertheless, there are still limited data on ethnic
differences in risk factors for DR. Understanding ethnic
differences in the prevalence and risk of DR will allow
better planning of public health measures and designing of
ethnicity-specific approaches to the management of DR.
Singapore, comprising people of Chinese, Malay, and
Indian origin, the 3 major ethnic groups in Asia, provides a

ª 2017 by the American Academy of Ophthalmology https://doi.org/10.1016/j.ophtha.2017.10.026 1

Published by Elsevier Inc. ISSN 0161-6420/17
 Ophthalmology Volume -, Number -, Month 2017
unique opportunity to examine possible ethnic differences in
DR in Asians. Thus, the aim of this study is to compare the
prevalence and risk factors of DR in the 3 main ethnic
groups in Singapore.
Methods
Study Design and Population
The Singapore Epidemiology of Eye Diseases (SEED,
2004e2011) Study is a population-based study that included 3
major ethnic groups in Singapore: Malays (2004e2006), Indians
(2007e2009), and Chinese (2009e2011).16,17 The study adhered
to the Declaration of Helsinki, and ethics approval was obtained
from the Singapore Eye Research Institute Institutional Review
Board. On the basis of an age-stratified random sampling strategy,
5000 Malays, 6350 Indians, and 6752 Chinese names were
selected. Of these, 4168 Malay, 4497 Indian, and 4605 Chinese
individuals were deemed eligible to participate. The “ineligible”
persons included those who had moved out from the residential
address, had not lived there in the past 6 months, or were deceased
or terminally ill. A total of 3280 Malays (response rate: 78.7%),
3400 Indians (75.6%), and 3353 Chinese (72.8%), aged 40 to 80þ
years, participated in this population-based study. The study aimed
to examine an approximately equal number of participants from
each ethnic group to provide robust comparable data across
ethnicity. All participants underwent standardized clinical and
ocular assessment, questionnaire interview, and blood biochemical
analyses.
Retinal Photography and Diabetic Retinopathy
Assessment
Two-field, 45-degree digital retinal photography was undertaken
using a standardized protocol.10,18 After pupil dilation, 1 retinal
photograph centered at the optic disc (Early Treatment Diabetic
Retinopathy Study [ETDRS] field 1) and another centered on the
macula (field 2) were taken from both eyes using a digital retinal
camera (Canon CR-DGi with a 10-D SLR back; Canon, Tokyo,
Japan). Photographs were sent and graded at the University of
Sydney by graders trained by the Blue Mountains Eye Study
principle investigator (P.M.), based on the modified Airlie House
Classification Scheme of the ETDRS.10,18,19 Diabetic retinopathy
was considered present if any characteristic lesion was present:
microaneurysms, hemorrhages, cotton wool spots, intraretinal
microvascular abnormalities, hard exudates, venous beading, and
new vessels. Severity of DR was graded according to a scale
modified from the Airlie House classification system: Retinopathy
severity was categorized into minimal nonproliferative DR
(NPDR) (levels 15e20), mild NPDR (level 35), moderate NPDR
(levels 43e47), severe NPDR (level 53), and proliferative DR
(PDR) (levels >60).
Diabetic macular edema was defined as hard exudates in the
presence of microaneurysms and blot hemorrhage within 1 disc
diameter from the foveal center or presence of focal photocoagu-
lation scars in the macular areas. Those with DME were further
divided into cases with clinically significant macular edema
(CSME), defined as DME within 500 mm of the foveal center or if
focal photocoagulation scars were present in the macular area.
Vision-threatening diabetic retinopathy (VTDR) was defined as the
presence of severe NPDR, PDR, or CSME using the Eye Diseases
Prevalence Research Group definition.20 For every participant, the
severity score of the worse eye was used for the analyses. If an eye
was ungradable, the score for the fellow eye was used to define
these outcomes.
2
Risk Factor Measurements and Definitions
All participants underwent a detailed interview: information on
socioeconomic status (education, income, housing status), lifestyle
risk factors (smoking and alcohol consumption), medication use,
and self-reported history of diabetes, hypertension, and cardio-
vascular disease (myocardial infarction, stroke) was collected.
Educational level was categorized into 4 groups (polytechnic/uni-
versity, secondary education, primary education, and no formal
education); monthly individual income level was categorized into 2
groups (<Singapore dollar [S$]<1000 and S$1000); and housing
type was categorized into 2 groups (4-room Housing Develop-
ment Board flat vs. 5-room Housing Development Board flat or
private apartment/house). Height was measured using a wall-
mounted tape and weight with a digital scale (SECA, model 782
2321009; Vogel & Halke, Hamburg, Germany). We defined body
mass index (BMI) as weight divided by the square of height in
meters (kg/m2). Blood pressure was measured using a digital
automatic blood pressure monitor (Dinamap model Pro100V2;
Criticon GmbH, Norderstedt, Germany), following the protocol
used in the Multi-Ethnic Study of Atherosclerosis.10 Hypertension
was defined as systolic blood pressure 140 mmHg, diastolic
blood pressure 90 mmHg, or use of antihypertensive
medication. Nonfasting venous blood samples were drawn and
sent for analysis of serum lipid levels (total cholesterol, high-
density lipoprotein cholesterol, and low-density lipoprotein
[LDL] cholesterol), hemoglobin A1C, creatinine, and glucose at
the National University Hospital Reference Laboratory on the same
day. Diabetes was defined as casual plasma glucose 200 mg/dl
(11.1 mmol/l), self-reported physician-diagnosed diabetes or gly-
cated hemoglobin A1c (HbA1c) >6.5%, and use of glucose-
lowering medication in accordance with the recommendation by
the American Diabetes Association.21 A participant was considered
to have type 1 diabetes if the participant was aged <30 years when
diagnosed with diabetes and was receiving insulin therapy with no
other hypoglycemic medication history. Hyperlipidemia was
defined as total cholesterol of 6.2 mmol/l or use of lipid-
lowering drugs. Chronic kidney disease (CKD) was defined as
an estimated glomerular filtration rate (eGFR) of <60 ml/min per
1.73 m2, measured from serum creatinine.22
Statistical Analysis
Analyses were performed using Stata version 14.2 (StataCorp LP,
College Station, TX) and R version 3.2.1 (Foundation for Statis-
tical Computing, Vienna, Austria). We calculated the overall and
ethnicity-specific prevalence of any DR, and DME, VTDR, and
CSME in those with diabetes. Age-standardized prevalence esti-
mates were calculated by the direct method using the 2010
Singapore population census as the standard population. Charac-
teristics of participants by any DR were compared using the t test
for continuous variables (e.g., age and BMI) and chi-square test or
Fisher exact test for categoric variables (e.g., hypertension and
ethnicity), as appropriate.
The associations of potential risk factors with the presence of
any DR in diabetic individuals as a whole, and within each ethnic
stratum, were analyzed in separate logistic regression models that
adjusted for (1) age and gender, and (2) additionally known
metabolic and socioeconomic risk factors of DR, including BMI,
HbA1c, systolic blood pressure, diabetes duration, income
(<S$2000), housing types (4 rooms or smaller), and education
levels (primary school or below). Where a variable was overall
found to be significantly associated with DR after multivariable
adjustment, an expanded model that included multiplicative pa-
rameters of the variable with ethnicity was used to test for signif-
icant interactions with ethnicity. Similar multivariable model
 Tan et al 
Ethnicity Is a Risk Factor for DR

adjustments were applied to identify risk factors independently
associated with any DME and VTDR, respectively. Last, to eval-
uate whether the higher prevalence of all 3 conditions in Indians
could be explained by additional cardiovascular and renal risk
factors, we examined the change in each odds ratio (OR) after
further adjusting for mean arterial blood pressure, antihypertension
medication use, total cholesterol, high-density lipoprotein choles-
terol, LDL cholesterol, serum glucose, smoking status, alcohol use,
myocardial infarction history, stroke history, and eGFR. All effect
estimates were presented with 95% confidence intervals (CIs), and
statistical significance was defined as P < 0.05 in 2-sided hy-
pothesis tests.
Results
Of the total of 10 033 persons who participated in SEED, 2964 had
diabetes. After excluding those with ungradable photographs
(n ¼ 87, 2.9%), 2877 persons with diabetes (1008 Malays, 1288
Indians, 581 Chinese) were included for the final analysis. Of these,
51 (1.7%) had type 1 diabetes and the rest had type 2 diabetes.
The age-standardized prevalence was 28.2% (95% CI,
25.9e30.6) for any DR, 7.6 (95% CI, 6.5e9.0) for any DME, 7.7%
(95% CI, 6.6e9.0) for VTDR, and 6.4% (95% CI, 5.4e7.6) for
CSME. Indians had a higher prevalence of any DR (30.7% vs.
26.2% in Chinese and 25.5% in Malays, P ¼ 0.012). Similar trends
were noted for any DME (9.5% vs. 6.1% in Chinese and 5.7% in
Malays, P ¼ 0.001) and CSME (7.6% vs. 5.9% in Chinese and
5.0% in Malays, P ¼ 0.032). There was no significant difference in
the prevalence of VTDR across the ethnic groups (P ¼ 0.078)
(Tables 1 and 2). Pairwise differences were observed between
Malays and Indians for all conditions and between Chinese and
Indians for DME only.
Table 3 shows characteristics of participants stratified by DR
status. Those with DR were more likely to have higher HbA1c,
higher systolic blood pressure, longer duration of diabetes,
higher serum total and LDL cholesterol, higher serum creatinine,
lower BMI, lower education, and lower income, and to live in
smaller housing; were more likely to have higher prevalence of
hypertension, previous myocardial infarct, and previous stroke,
and less likely to be current smokers compared with those
without DR (all P < 0.05).
In multivariable analysis including potential risk factors
(Table 4), younger age, Indian ethnicity, longer diabetes duration,
higher HbA1c, higher serum glucose, higher systolic blood
pressure, lower diastolic blood pressure, lower total cholesterol
and LDL cholesterol, CKD, and lower education were
independently associated with higher odds of any DR.
Vision-threatening diabetic retinopathy showed similar risk
factors to DR and was associated with a previous stroke and
smoking but not with total and LDL cholesterol, education, or
ethnicity. Likewise, DME showed generally similar risk factors to
that for any DR and was associated with lower high-density lipo-
protein, higher serum triglycerides, and history of stroke, but not
associated with total and LDL cholesterol, or education (Table 4).
We evaluated the factors that may explain the higher prevalence
of DR among Indians (Table 5). We found that OR of any DR for
Indians relative to Chinese (OR of 1.42, Model 1) increased by
11% after adjusting for DR risk factors, vascular risk factors, and
eGFR (adjusted OR of 1.53, Model 3). A similar trend of
increased odds for DME and VTDR was found after adjusting
for known risk factors and eGFR.
Table 6 shows the major risk factors for DR stratified by ethnic
groups. In general, risk factors were similar across the 3 ethnic
groups, including higher HbA1c, longer duration of diabetes,
higher systolic blood pressure, lower diastolic blood pressure,
and CKD. Lower total (P ¼ 0.001) and LDL cholesterol (P ¼
0.001) were associated with any DR in Malays only. Increasing
age was associated with lower odds of DR in Chinese (P <
0.001) and Indians (P ¼ 0.024) but not in Malays (P ¼ 0.058).
A history of stroke was associated with VTDR in Indians only.
There was no interaction among ethnicity with HbA1c, duration
of diabetes, blood pressure, or lipids for the risk of DR. The
only significant interaction was found between diabetes duration
and age. In a sensitivity analysis performed using the full
categoric information available for socioeconomic status
variables, increasing age was associated with lower risk of DR
for all 3 ethnic groups and hypertension was associated with
increasing risk of DR. All other results remained unchanged.
Discussion
In this study of a multi-ethnic Asian population aged 40 to
80þ years with diabetes, we showed that 1 in 3 had DR, and 1
in 10 had vision-threatening levels of DR. The lower preva-
lence of diabetes in Chinese is consistent with the findings of

Table 1. Prevalence of Diabetic Retinopathy, Diabetic Macula Edema, and Vision-Threatening Diabetic Retinopathy by Race*

All (n [ 2877) Chinese (n [ 581) Malay (n [ 1008) Indian (n [ 1288)

%* 95% CI %* 95% CI %* 95% CI %* 95% CI P Valuey
Any DR 28.2 25.9e30.6 26.2 20.7e33.1 25.5 21.8e30.3 30.7 27.5e34.4 0.012
Minimal DR 9.83 8.53e11.3 8.43 5.56e12.7 9.76 7.62e13.2 10.4 8.51e12.6
Mild DR 7.85 6.63e9.28 5.99 3.43e10.1 4.91 3.27e8.02 10.6 8.67e12.9
Moderate DR 6.02 4.97e7.27 7.77 4.70e12.4 6.09 4.40e9.20 5.37 4.02e7.10
Severe DR 0.71 0.37e1.29 0.60 0.02e3.37 1.29 0.52e3.88 0.40 0.10e1.19
Proliferative DR 3.75 3.04e4.65 3.39 2.18e5.99 3.43 2.25e6.16 4.00 2.96e5.39
Any DME 7.62 6.46e8.97 6.05 3.60e10.0 5.67 4.11e8.65 9.49 7.68e11.7 0.001
CSME 6.39 5.37e7.60 5.88 3.45e9.83 4.96 3.61e7.76 7.59 6.03e9.52 0.032
VTDR 7.68 6.55e9.00 6.93 4.37e10.9 6.44 4.74e9.53 8.87 7.18e10.9 0.078

CI ¼ confidence interval; CSME ¼ clinically significant macular edema; DME ¼ diabetic macular edema; DR ¼ diabetic retinopathy; VTDR ¼ vision-
threatening diabetic retinopathy.
Boldface indicates statistical significance.
*Prevalence rates were standardized to the 2010 Singapore Resident Population.
y
Chi-square test P value comparing age-standardized prevalence between ethnic groups.

3
 Ophthalmology Volume -, Number -, Month 2017

Table 2. Pairwise Comparison of the Age-Standardized Chinese and Malays. However, major risk factors for DR
Prevalence between Ethnic Groups* were similar across the 3 ethnic groups.
There is little available data from Asia directly comparing
Chinese vs. Malay Chinese vs. Indian Malay vs. Indian the prevalence of DR between ethnic groups. Few previous
Any DR 0.803 0.052 0.006 studies that compared prevalence of DR between Western
Any DME 0.835 0.017 0.001 and Asian populations in western countries have provided
CSME 0.501 0.215 0.014 indirect evidence to support the higher prevalence of DR in
VTDR 0.785 0.186 0.038 Asians compared with whites.25,26 Studies in the United
States have noted that Chinese and South Asian patients had a
CSME ¼ clinically significant macular edema; DME ¼ diabetic macular higher prevalence of DR than whites,27 and that the
edema; DR ¼ diabetic retinopathy; VTDR ¼ vision-threatening diabetic prevalence of DR and DME was significantly higher in
retinopathy.
Boldface indicates statistical significance. blacks and Hispanics compared with whites and Chinese
*Chi-square test P value. subjects.10 A pooled analysis from several population-
based studies found the prevalence of DR varied among
ethnic groups and was the highest in African Americans and
2 previous studies. The Singapore National Health Survey lowest in Asians.8 Our study found that Indian ethnicity was
found that the prevalence of diabetes in adults aged 18 to 69 associated with higher odds of DR, compared with Chinese
years was 15.3% in Indians, 11.0% in Malays, and 7.1% in and Malays. In this regard, 2 previous studies also reported
Chinese.23 Likewise, in the Singapore Prospective Study that Indian/South Asian ethnicity was an independent risk
Program, among an older group of participants (40e95 factor for DR; however, these were compared against a
years), the overall age-standardized prevalence of diabetes white population.9,27 In the United States, Hispanic
was 21.6% in Indians, 17.1% in Malays, and 11.5% in ethnicity was found to be an independent risk factor for
Chinese.24 Our study found that the prevalence of any DR DR.14,28 Because the socioeconomic status and access to care
and any DME was higher in Indians compared with vary among ethnic groups in these studies, we opine that our

Table 3. Characteristic of Population by Presence of Diabetic Retinopathy

No DR DR
Factor Mean (SD) or % (n) Mean (SD) or % (n) P Valuey
Age, yrs 61.5 (10.2) 61.9 (9.20) 0.335
Male* 51.1% (1029) 49.8% (428) 0.567
Ethnicity
Chinese 432 (21.4) 149 (17.3) 0.009
Malay 716 (35.5) 292 (34.0)
Indian 870 (43.1) 418 (48.7)
BMI, kg/m2 27.1 (4.84) 26.4 (4.68) <0.001
HbA1c, % 7.49 (1.54) 8.26 (1.85) <0.001
Diabetes duration (per year) 5.47 (7.68) 12.3 (10.1) <0.001
Glucose, mmol/l 9.27 (4.20) 11.2 (5.73) <0.001
Systolic BP, mmHg 143.5 (20.8) 149.7 (23.5) <0.001
Diastolic BP, mmHg 78.3 (10.2) 77.8 (11.2) 0.237
Hypertension* 78.3% (1577) 86.2% (740) <0.001
Total cholesterol, mmol/l 5.20 (1.21) 5.07 (1.32) 0.015
HDL cholesterol, mmol/l 1.14 (0.32) 1.16 (0.35) 0.111
LDL cholesterol, mmol/l 3.21 (1.00) 3.05 (1.03) <0.001
Triglycerides, mmol/l 2.04 (1.38) 2.05 (1.47) 0.923
Previous MI* 11.0% (221) 14.1% (121) 0.017
Previous stroke* 3.7% (75) 6.3% (54) 0.002
Current smoker* 14.5% (293) 11.7% (100) 0.039
Education (secondary or higher)* 33.4% (673) 24.9% (213) <0.001
Income (<S$2000/mo)* 82.4% (1632) 86.8% (732) 0.004
Housing type (4 rooms)* 71.9% (1449) 77.4% (664) 0.002
Creatinine (per mmol/l) 83.9 (42.7) 98.4 (75.6) <0.001
eGFR, mL/min/1.73 m2/yr 80.7 (21.1) 74.9 (26.6) <0.001
CKD 16.7% (327) 28.3% (231) <0.001

BMI ¼ body mass index; BP ¼ blood pressure; CKD ¼ chronic kidney disease; DR ¼ diabetic retinopathy; eGFR ¼ estimated glomerular filtration rate;
HbA1c ¼ glycated hemoglobin A1c; HDL ¼ high-density lipoprotein; LDL ¼ low-density lipoprotein; MI ¼ myocardial infarction; SD ¼ standard
deviation.
Boldface indicates statistical significance.
*Chi-square/Fisher exact test.
y
P value with t test assuming equal variance.

4
 Tan et al 
Ethnicity Is a Risk Factor for DR

Table 4. Risk Factors for Diabetic Macular Edema and Vision-Threatening Diabetic Retinopathy

DR DME VTDR
Multivariable Adjusted* Multivariable Adjusted* Multivariable Adjusted*
Risk Factor OR (95% CI) P Value OR (95% CI) P Value OR (95% CI) P Value
Age (per 10-yr increase) 0.77 (0.68e0.86) <0.001 0.69 (0.57e0.84) <0.001 0.73 (0.60e0.88) 0.001
Gender (female) 0.91 (0.74e1.10) 0.325 1.06 (0.76e1.49) 0.714 1.10 (0.79e1.53) 0.577
Ethnicity: Chinese 1.0 1.0 1.0
Malay 0.97 (0.73e1.29) 0.845 1.06 (0.62e1.79) 0.838 0.85 (0.52e1.39) 0.522
Indian 1.41 (1.09e1.83) 0.009 1.89 (1.17e3.05) 0.009 1.46 (0.94e2.29) 0.093
BMI (kg/m2) 0.98 (0.96e1.00) 0.076 0.97 (0.94e1.01) 0.124 0.97 (0.94e1.01) 0.137
HbA1c (per mmol/l) 1.25 (1.18e1.32) <0.001 1.29 (1.19e1.40) <0.001 1.26 (1.17e1.37) <0.001
Diabetes duration (per year) 1.10 (1.08e1.11) <0.001 1.09 (1.07e1.10) <0.001 1.10 (1.08e1.12) <0.001
Serum glucose (per mmol/l) 1.03 (1.00e1.06) 0.021 1.07 (1.03e1.11) <0.001 1.06 (1.02e1.10) 0.003
Systolic BP (per 10 mmHg, increase?) 1.14 (1.09e1.19) <0.001 1.14 (1.06e1.22) <0.001 1.16 (1.08e1.24) <0.001
Diastolic BP (per 10 mmHg) 0.74 (0.65e0.84) <0.001 0.61 (0.49e0.77) <0.001 0.57 (0.46e0.72) <0.001
Hypertension 1.33 (1.00e1.77) 0.053 1.26 (0.76e2.11) 0.371 1.72 (0.99e2.97) 0.054
Total cholesterol (per mmol/l, increase?) 0.87 (0.80e0.95) 0.002 0.97 (0.84e1.11) 0.657 1.00 (0.88e1.15) 0.961
HDL cholesterol (per mmol/l) 1.01 (0.75e1.36) 0.965 0.54 (0.33e0.91) 0.019 0.73 (0.45e1.18) 0.201
LDL cholesterol (per mmol/l) 0.83 (0.74e0.92) <0.001 0.87 (0.73e1.03) 0.108 0.93 (0.78e1.09) 0.364
Triglycerides (per mmol/l) 0.98 (0.92e1.05) 0.584 1.13 (1.02e1.24) 0.015 1.09 (0.99e1.20) 0.097
Current smoker 0.79 (0.58e1.06) 0.117 0.65 (0.37e1.14) 0.134 0.53 (0.29e0.97) 0.039
Education: primary or below 1.47 (1.16e1.86) 0.001 1.28 (0.85e1.93) 0.230 1.23 (0.82e1.84) 0.313
Income: <S$2000 0.89 (0.66e1.20) 0.441 1.28 (0.73e2.25) 0.387 1.33 (0.75e2.37) 0.331
Housing: 3e4 rooms or less 1.17 (0.93e1.47) 0.183 1.12 (0.75e1.66) 0.574 1.27 (0.85e1.90) 0.236
Previous myocardial infarct 1.12 (0.84e1.51) 0.435 1.37 (0.87e2.15) 0.170 1.48 (0.96e2.28) 0.074
Previous stroke 1.55 (0.99e2.43) 0.056 2.42 (1.33e4.40) 0.004 3.27 (1.87e5.72) <0.001
Serum creatinine (per mmol/l) 1.00 (1.00e1.01) <0.001 1.01 (1.00e1.01) <0.001 1.01 (1.01e1.01) <0.001
eGFR 0.99 (0.98e0.99) <0.001 0.98 (0.97e0.99) <0.001 0.97 (0.97e0.98) <0.001
CKD 1.81 (1.42e2.30) <0.001 3.21 (2.23e4.62) <0.001 4.33 (3.02e6.19) <0.001

BMI ¼ body mass index; BP ¼ blood pressure; CI ¼ confidence interval; CKD ¼ chronic kidney disease; DME ¼ diabetic macular edema; DR ¼ diabetic
retinopathy; eGFR ¼ estimated glomerular filtration rate; HbA1c ¼ glycated hemoglobinA1c; HDL ¼ high-density lipoprotein; LDL ¼ low-density li-
poprotein; OR ¼ odds ratio; VTDR ¼ vision-threatening diabetic retinopathy.
*Adjusted for age, gender, BMI, HbA1c, SBP, duration of diabetes, income, housing, and education level.

study, which consisted of 3 Asian ethnic groups residing
within a similar environment with good access to public
healthcare, may more accurately reflect the ethnic differences
in this aspect.
We found that glycosylated hemoglobin, duration of
diabetes, and increased systolic blood pressure, the major
risk factors for DR, to be similar across the 3 ethnic groups.
This is consistent with prior epidemiology studies on DR.29
Increasing age may be associated with a reduced prevalence
of DR because of the competing risk of death in the older
population. Serum cholesterol was found to be associated
with DR in the Malay population only, with lower levels
of total and LDL cholesterol to be associated with higher
risk. Prior epidemiologic studies did not report a
consistent association between serum lipids and DR.30
Treatment with statins as a class of lipid-lowering medica-
tions was not associated with DR in this population, but
lipid treatment guidelines that recommend more aggressive
lipid lowering for those with higher risk of vascular com-
plications may account for the lower lipid levels in those
with DR.31 Current smokers were found to have a lower risk
of VTDR, which may be due to selective mortality of those

Table 5. Effect of Additional Factors on the Odds Ratio of Diabetic Retinopathy, Diabetic Macular Edema, and Vision-Threatening
Diabetic Retinopathy in Indians Compared with Chinese

OR in Indians with Reference to Chinese (95% CI)

Adjustment (Models) DR DME VTDR
Age, gender (1) 1.42 (1.14e1.77) 1.75 (1.18e2.60) 1.42 (0.99e2.05)
Plus known risk factors (2) 1.41 (1.09e1.83) 1.89 (1.17e3.05) 1.46 (0.94e2.29)
Plus additional vascular risk factors and renal disease (3) 1.53 (1.16e2.01) 2.01 (1.21e3.33) 1.54 (0.96e2.48)

CI ¼ confidence interval; DME ¼ diabetic macular edema; DR ¼ diabetic retinopathy; OR ¼ odds ratio; VTDR ¼ vision-threatening diabetic retinopathy.
Model 1: age and gender. Model 2: Model 1 variables plus BMI, HbA1c, SBP, duration of diabetes, income, housing, and education level. Model 3: Model 2
variables plus mean arterial pressure, antihypertension medication use, total cholesterol, high-density lipoprotein cholesterol, LDL cholesterol, serum
glucose, smoking status, alcohol use, myocardial infarction history, stroke history, and eGFR.

5
 Ophthalmology Volume -, Number -, Month 2017

Table 6. Risk Factors for Diabetic Retinopathy by Ethnicity

Chinese Malay Indian

Multivariable Adjusted* Multivariable Adjusted* Multivariable Adjusted*
Risk Factor OR (95% CI) for DR P Value OR (95% CI) for DR P Value OR (95% CI) for DR P Value
Age (per 10 yrs) 0.58 (0.44e0.77) <0.001 0.82 (0.68e1.01) 0.058 0.82 (0.69e0.97) 0.024
Gender (female) 0.88 (0.55e1.41) 0.602 1.13 (0.80e1.60) 0.487 0.77 (0.58e1.04) 0.088
BMI (kg/m2) 0.99 (0.93e1.05) 0.781 0.98 (0.94e1.01) 0.220 0.98 (0.96e1.01) 0.294
HbA1c (per mmol/l) 1.23 (1.05e1.44) 0.010 1.28 (1.18e1.40) <0.001 1.25 (1.15e1.37) <0.001
Diabetes duration (per year) 1.09 (1.06e1.12) <0.001 1.09 (1.07e1.11) <0.001 1.10 (1.08e1.12) <0.001
Serum glucose (per mmol/l) 1.04 (0.98e1.10) 0.205 1.03 (0.98e1.07) 0.245 1.03 (0.98e1.07) 0.225
Systolic BP (per 10 mmHg) 1.24 (1.10e1.40) <0.001 1.20 (1.11e1.28) <0.001 1.08 (1.01e1.16) 0.030
Diastolic BP (per 10 mmHg) 0.69 (0.49e0.98) 0.037 0.72 (0.58e0.89) 0.003 0.74 (0.61e0.90) 0.002
Hypertension 1.46 (0.66e3.23) 0.346 1.19 (0.67e2.10) 0.553 1.42 (0.98e2.06) 0.067
Total cholesterol (per mmol/l) 0.86 (0.69e1.07) 0.187 0.80 (0.69e0.92) 0.002 0.97 (0.85e1.11) 0.669
HDL cholesterol (per mmol/l) 0.97 (0.48e1.93) 0.923 1.24 (0.69e2.20) 0.471 1.44 (0.90e2.30) 0.126
LDL cholesterol (per mmol/l) 0.81 (0.62e1.06) 0.123 0.75 (0.64e0.89) 0.001 0.90 (0.77e1.06) 0.200
Triglycerides (per mmol/l) 0.94 (0.81e1.10) 0.449 0.95 (0.86e1.06) 0.382 0.98 (0.88e1.10) 0.768
Current smoker 0.58 (0.28e1.20) 0.142 1.05 (0.63e1.75) 0.840 0.78 (0.50e1.22) 0.280
Education: primary or below 2.43 (1.38e4.26) 0.002 1.85 (1.13e3.02) 0.014 1.15 (0.84e1.58) 0.381
Income: <S$2000 1.18 (0.59e2.36) 0.640 0.66 (0.34e1.26) 0.207 1.00 (0.67e1.48) 0.983
Housing: 3e4 rooms or less 1.33 (0.78e2.27) 0.287 1.15 (0.71e1.88) 0.567 1.24 (0.91e1.69) 0.173
Previous myocardial infarct 1.09 (0.50e2.37) 0.832 1.51 (0.89e2.55) 0.127 0.89 (0.59e1.33) 0.564
Previous stroke 1.06 (0.31e3.62) 0.920 1.14 (0.52e2.50) 0.753 2.15 (1.13e4.09) 0.019
Creatinine (per mmol/l) 1.01 (1.00e1.01) 0.008 1.00 (1.00e1.01) 0.028 1.01 (1.00e1.01) <0.001
eGFR 0.98 (0.97e0.99) 0.004 0.99 (0.98e1.00) 0.019 0.98 (0.98e0.99) <0.001
CKD 2.36 (1.22e4.54) 0.010 1.58 (1.09e2.29) 0.016 2.64 (1.75e4.00) <0.001

BMI ¼ body mass index; BP ¼ blood pressure; CI ¼ confidence interval; CKD ¼ chronic kidney disease; DR ¼ diabetic retinopathy; eGFR ¼ estimated
glomerular filtration rate; HbA1c ¼ glycated hemoglobin A1c; HDL ¼ high-density lipoprotein; LDL ¼ low-density lipoprotein; OR ¼ odds ratio.
*Adjusted for age, gender, BMI, HbA1c, SBP, duration of diabetes, income, housing, and education level.

with current smoking and VTDR. Although previous stroke
was significantly associated with DR in Indians, adjusting
for stroke did not decrease the higher OR for DR in
Indians. We also found a higher prevalence of
cardiovascular disease in Indians (21.7% vs. 17.8%
Malays; 14.3% Chinese). We did not find any other
significant risk factor in the Indian population that may
account for the excess risk of DR. Furthermore, adjusting
for known risk factors for retinopathy and vascular risk
factors, and CKD, results in an increase in the risk of DR,
DME, and VTDR associated with Indian ethnicity
(Table 5). However, there may be other environmental
factors such as diet, which may differ between ethnic
groups, for which we did not have data available.
There was no interaction found between ethnicity and
other risk factors for DR, including glycosylated hemoglobin,
duration of diabetes, blood pressure, or lipids. Raymond et al9
and Stolk et al,27 who reported an independent increased risk
of DR in South Asians compared with whites, found an
interaction between ethnicity and duration of diabetes.
Socioeconomic factors and access to medical care may
affect the relationship between ethnicity and duration of
diabetes in those populations. Our population had similar
good access to care, and although there were socioeconomic
differences with Malays having lower levels of education
and income, these factors were taken into account in the
multivariable analysis. The lack of interaction between
ethnicity and HbA1c on DR is consistent with that reported
in a prior study that compared whites, blacks, and Hispanic
ethnic groups.15 Our finding supports the use of a standard
HbA1c cutoff value in the diagnosis and screening of
diabetes across Asian ethnic groups.15,32,33
Given there is no difference in the risk factors for DR
between ethnic groups and no interaction between the
known major risk factors and ethnicity, genetic difference
between ethnic groups may account for some of these dif-
ferences in prevalence. There is evidence to support a ge-
netic component for DR.34,35 Familial aggregation studies
and clinical trials, including the Diabetes Control and
Complications Trial, suggested a heritable tendency for DR
independent of shared risk factors.36 Few genes have been
found be in involved in the pathogenesis of DR. The use
of genome-wide association analysis and gene sequencing
methodologies in appropriate populations with diabetes may
increase the likelihood of discovering genes and combina-
tions of genes that are closely related to the pathogenesis
of DR in the near future. Genetic association studies
performed on the SEED study population found type 2
diabeteseassociated genetic loci to be associated with
increased risk of DR, although no ethnic difference in ge-
netic risk at these loci was found.37 Further research needs
to be performed to further elucidate the cause for the
ethnic differences in prevalence and risk of DR.
The strengths of this work include a large population-
based, multiethnic sample, with standardized grading of
retinal photographs and detailed standardized assessment of
other systemic and ocular factors. However, a few limitations
should be acknowledged. Diabetic retinopathy was graded on

6
 Tan et al 
Ethnicity Is a Risk Factor for DR
2-field nonstereoscopic fundus photography, which may miss
peripheral lesions compared with standard 7-field ETDRS
photography, and thus potentially underestimating the
prevalence of DR. The DME criteria adopted in our study
may not be directly comparable to other studies using ste-
reoscopic photography or examination. Because of the
cross-sectional design of our study, causal relationships
between risk factors and DR may not be definitively
concluded. Although our study population encompasses the
3 major ethnic groups in Asia, the social, health, and living
conditions in Singapore are different from other Asian
countries. Thus, our findings may not be entirely extrapo-
lated to other Asian countries with the same ethnicity
group. Furthermore, data collected from each ethnic group
were over a time period from 2004 to 2011; diabetes care
may have changed over this 8-year period, which may
affect the prevalence of complications such as DR. By
comparing data from the Singapore National Health Survey
performed in 200423 and 2010,38 the age-standardized
prevalence of diabetes increased from 9.0% in 2004 to
11.3% in 2010. The survey also showed that 32.0% of
known persons with diabetes had poor blood glucose
control in 2010, higher than the proportion of 20.4% in
2004. If the ethnic differences in DR prevalence were due
to the time of data collection, based on the result from this
National health survey, the Chinese cohort (surveyed in
2010) would have had a higher prevalence of DR than the
early cohorts.
In conclusion, our study provides major population-
based data comparing the characteristics of DR among
Chinese, Malays, and Indians, the 3 major ethnic groups in
the whole of Asia and living in a homogenous environment,
for which there was limited prior data. In this large cohort,
we showed that 1 in 3 persons with diabetes had DR and 1
in 10 had VTDR. Although Indians had a higher prevalence
of any DR and DME, major DR risk factors were similar
across the 3 ethnic groups. Asia has a rapidly increasing
population with diabetes.5e7 Given that guidelines on DR
care are variable in Asia,39 our results suggest that a more
common and standardized approach in addressing DR risk
factors, in particular blood glucose and blood pressure
control, may reduce the public health impact and burden
of this preventable cause of vision loss in Asia.
References
1. Klein BE. Overview of epidemiologic studies of diabetic
retinopathy. Ophthalmic Epidemiol. 2007;14:179-183.
2. Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lan-
cet. 2010;376:124-136.
3. Wong TY, Cheung CM, Larsen M, et al. Diabetic retinopathy.
Nat Rev Dis Primers. 2016;2:16012.
4. Guariguata L, Whiting DR, Hambleton I, et al. Global esti-
mates of diabetes prevalence for 2013 and projections for
2035. Diabetes Res Clin Pract. 2014;103:137-149.
5. Chan JC, Malik V, Jia W, et al. Diabetes in Asia: epidemi-
ology, risk factors, and pathophysiology. JAMA. 2009;301:
2129-2140.
6. Xu Y, Wang L, He J, et al. Prevalence and control of diabetes
in Chinese adults. JAMA. 2013;310:948-959.
7. Sabanayagam C, Yip W, Ting DS, et al. Ten emerging trends
in the epidemiology of diabetic retinopathy. Ophthalmic Epi-
demiol. 2016;23:209-222.
8. Yau JW, Rogers SL, Kawasaki R, et al. Global prevalence and
major risk factors of diabetic retinopathy. Diabetes Care.
2012;35:556-564.
9. Raymond NT, Varadhan L, Reynold DR, et al. Higher prev-
alence of retinopathy in diabetic patients of South Asian
ethnicity compared with white Europeans in the community: a
cross-sectional study. Diabetes Care. 2009;32:410-415.
10. Wong TY, Klein R, Islam FM, et al. Diabetic retinopathy in a
multi-ethnic cohort in the United States. Am J Ophthalmol.
2006;141:446-455.
11. Sivaprasad S, Gupta B, Crosby-Nwaobi R, Evans J. Preva-
lence of diabetic retinopathy in various ethnic groups: a
worldwide perspective. Surv Ophthalmol. 2012;57:347-370.
12. Varma R, Bressler NM, Doan QV, et al. Prevalence of and risk
factors for diabetic macular edema in the United States. JAMA
Ophthalmol. 2014;132:1334-1340.
13. Klein R, Sharrett AR, Klein BE, et al. The association of
atherosclerosis, vascular risk factors, and retinopathy in adults
with diabetes: the Atherosclerosis Risk in Communities study.
Ophthalmology. 2002;109:1225-1234.
14. Harris MI, Klein R, Cowie CC, et al. Is the risk of diabetic
retinopathy greater in non-Hispanic blacks and Mexican
Americans than in non-Hispanic whites with type 2 diabetes?
A U.S. population study. Diabetes Care. 1998;21:1230-1235.
15. Bower JK, Brancati FL, Selvin E. No ethnic differences in the
association of glycated hemoglobin with retinopathy: the Na-
tional Health and Nutrition Examination Survey 2005-2008.
Diabetes Care. 2013;36:569-573.
16. Lavanya R, Jeganathan VS, Zheng Y, et al. Methodology of
the Singapore Indian Chinese Cohort (SICC) eye study:
quantifying ethnic variations in the epidemiology of eye dis-
eases in Asians. Ophthalmic Epidemiol. 2009;16:325-336.
17. Foong AW, Saw SM, Loo JL, et al. Rationale and method-
ology for a population-based study of eye diseases in Malay
people: the Singapore Malay Eye Study (SiMES). Ophthalmic
Epidemiol. 2007;14:25-35.
18. Wong TY, Cheung N, Tay WT, et al. Prevalence and risk
factors for diabetic retinopathy: the Singapore Malay Eye
Study. Ophthalmology. 2008;115:1869-1875.
19. Mitchell P, Smith W, Wang JJ, Attebo K. Prevalence of dia-
betic retinopathy in an older community. The Blue Mountains
Eye Study. Ophthalmology. 1998;105:406-411.
20. Kempen JH, O’Colmain BJ, Leske MC, et al. The prevalence
of diabetic retinopathy among adults in the United States. Arch
Ophthalmol. 2004;122:552-563.
21. American Diabetes Association. Standards of medical care in
diabetese2010. Diabetes Care. 2010;33(Suppl 1):S11-S61.
22. Shankar A, Leng C, Chia KS, et al. Association between body
mass index and chronic kidney disease in men and women:
population-based study of Malay adults in Singapore. Nephrol
Dial Transplant. 2008;23:1910-1918.
23. Singapore National Health Survey 2004. [article online]. Min-
istry of Health, Singapore 2005. Available at: http://www.moh.
gov.sg/content/moh_web/home/Publications/Reports/2005/
national_health_survey2004.html. Accessed August 2, 2017.
24. Chiang PP, Lamoureux EL, Cheung CY, et al. Racial differ-
ences in the prevalence of diabetes but not diabetic retinopathy
in a multi-ethnic Asian population. Invest Ophthalmol Vis Sci.
2011;52:7586-7592.
25. Sivaprasad S, Gupta B, Gulliford MC, et al. Ethnic variations
in the prevalence of diabetic retinopathy in people with dia-
betes attending screening in the United Kingdom (DRIVE
UK). PLoS One. 2012;7:e32182.
7
 Ophthalmology Volume -, Number -, Month 2017
26. Kliner M, Fell G, Gibbons C, et al. Diabetic retinopathy equity
profile in a multi-ethnic, deprived population in Northern
England. Eye (Lond). 2012;26:671-677.
27. Stolk RP, van Schooneveld MJ, Cruickshank JK, et al. Retinal
vascular lesions in patients of Caucasian and Asian origin with
type 2 diabetes: baseline results from the ADVANCE Retinal
Measurements (AdRem) study. Diabetes Care. 2008;31:708-713.
28. Emanuele N, Sacks J, Klein R, et al. Ethnicity, race, and
baseline retinopathy correlates in the veterans affairs diabetes
trial. Diabetes Care. 2005;28:1954-1958.
29. Lee R, Wong TY, Sabanayagam C. Epidemiology of diabetic
retinopathy, diabetic macular edema and related vision loss.
Eye Vis (Lond). 2015;2:17.
30. Lim LS, Wong TY. Lipids and diabetic retinopathy. Expert
Opin Biol Ther. 2012;12:93-105.
31. Jellinger PS, Handelsman Y, Rosenblit PD, et al. American
Association of Clinical Endocrinologists and American Col-
lege of Endocrinology Guidelines for Management of Dysli-
pidemia and Prevention of Cardiovascular Disease. Endocr
Pract. 2017;23(Suppl 2):1-87.
32. Sabanayagam C, Liew G, Tai ES, et al. Relationship between
glycated haemoglobin and microvascular complications: is
there a natural cut-off point for the diagnosis of diabetes?
Diabetologia. 2009;52:1279-1289.
Footnotes and Financial Disclosures
Originally received: June 15, 2017.
Final revision: September 28, 2017.
Accepted: October 17, 2017.
Available online: ---. Manuscript no. 2017-1382.
1
Singapore Eye Research Institute, Singapore National Eye Centre,
Singapore.
2 Obtained funding: Not applicable
Department of Ophthalmology, Yong Loo Lin School of Medicine, Na-
tional University of Singapore and National University Health System,
Singapore.
3
Duke-National University of Singapore Medical School, Singapore.
4
Department of Ophthalmology and Visual Sciences, Khoo Teck Phuat
Hospital, Singapore.
5 DME ¼ diabetic macular edema; DR ¼ diabetic retinopathy;
Center for Vision Research, University of Sydney, Sydney, Australia.
Financial Disclosure(s):
The author(s) have no proprietary or commercial interest in any materials
discussed in this article.
Funded by the National Medical Research Council (grants 0796/2003,
IRG07nov013, IRG09nov014, STaR/0003/2008; CG/SERI/2010) and
Biomedical Research Council (grants 08/1/35/19/550, 09/1/35/19/616),
Singapore. The sponsor or funding organization had no role in the design or
conduct of this research.
HUMAN SUBJECTS: Study protocol was approved by IRB/ethics com-
mittee of the Singapore Eye Research Institute Institutional Review Board.
All studies adhered to the tenets of the Declaration of Helsinki.
33. Tsugawa Y, Mukamal KJ, Davis RB, et al. Should the he-
moglobin A1c diagnostic cutoff differ between blacks and
whites? A cross-sectional study. Ann Intern Med. 2012;157:
153-159.
34. Liew G, Klein R, Wong TY. The role of genetics in suscep-
tibility to diabetic retinopathy. Int Ophthalmol Clin. 2009;49:
35-52.
35. Abhary S, Hewitt A, Burdon K, Craig J. A systematic meta-
analysis of genetic association studies for diabetic retinop-
athy. Diabetes. 2009;58:2137-2147.
36. Clustering of long-term complications in families with diabetes
in the diabetes control and complications trial. The Diabetes
Control and Complications Trial Research Group. Diabetes.
1997;46:1829-1839.
37. Chong YH, Fan Q, Tham YC, et al. Type 2 diabetes genetic
variants and risk of diabetic retinopathy. Ophthalmology.
2017;124:336-342.
38. Singapore National Health Survey 2010. [article online]. Min-
istry of Health, Singapore 2011. Available at: https://www.moh.
gov.sg/content/moh_web/home/Publications/Reports/2011/
national_health_survey2010.html. Accessed August 2, 2017.
39. Wang LZ, Cheung CY, Tapp RJ, et al. Availability and vari-
ability in guidelines on diabetic retinopathy screening in Asian
countries. Br J Ophthalmol. 2017;101:1352-1360.
Author Contributions:
Conception and design: Tan, Gan, Sabanayagam, Mitchell, Wang,
Lamoureux, Cheng, Wong
Data collection: Tan, Gan, Sabanayagam, Tham, Neelam, Mitchell, Cheng
Analysis and interpretation: Tan, Gan, Sabanayagam, Tham, Cheng, Wong
Overall responsibility: Tan, Gan, Sabanayagam, Tham, Neelam, Mitchell,
Wang, Lamoureux, Cheng, Wong
Abbreviations and Acronyms:
BMI ¼ body mass index; CI ¼ confidence interval; CKD ¼ chronic
kidney disease; CSME ¼ clinically significant macular edema;
eGFR ¼ estimated glomerular filtration rate; ETDRS ¼ Early Treatment
Diabetic Retinopathy Study; HbA1c ¼ glycated hemoglobin A1c;
LDL ¼ low-density lipoprotein; NPDR ¼ nonproliferative diabetic reti-
nopathy; OR ¼ odds ratio; PDR ¼ proliferative diabetic retinopathy;
SEED ¼ Singapore Epidemiology of Eye Diseases; VTDR ¼ vision-
threatening diabetic retinopathy.
Correspondence:
Tien Y. Wong, MD, PhD, Singapore Eye Research Institute, Singapore
National Eye Center, 11 Third Hospital Avenue, Singapore 168751. E-mail:
wong.tien.yin@snec.com.sg.